大数据处理警示性图像颜色纹理特征选取仿真  

Big Data Processing Warning Image Color Texture Feature Selection Simulation

作　　者：张子栋[1] 张杰敏[1] 茅剑[1] ZHANG Zi-dong;ZHANG Jie-min;MAO Jian(Computer Engineering College,Jimei University,Fujian Xiamen 361021,China)

机构地区：[1]集美大学计算机工程学院

出　　处：《计算机仿真》2019年第5期434-437,470,共5页Computer Simulation

基　　金：福建省自然科学基金(2017J01762);福建省教育厅(JA15278)

摘　　要：警示性图像颜色纹理特征选取的有效性直接影响着多个领域的发展。针对当前方法存在的警示性图像颜色纹理特征选取准确率低的问题,提出了一种基于维纳滤波和增强矩阵的警示性图像颜色纹理特征选取方法,采用维纳滤波对警示性图像中的噪声进行消除,噪声消除后利用横向增量与纵向增量计算警示性图像中像素点的灰度值,求解每个像素点灰度值所对应的横纵坐标轴的二阶偏导数之和,根据这个偏导数推导出警示性图像的颜色纹理特征矩阵,利用该矩阵得到了一个警示性图像的颜色纹理特征模型,通过这个模型实现了对大数据处理下警示性图像颜色纹理特征的选取。仿真结果表明,所提方法对警示性图像的噪声具有很好的抑制效果,可以较好的保留警示性图像中的信息,并且所提方法能够准确的在大数据处理环境下对警示性图像颜色纹理特征的选取。Current feature selection method of color texture of warning image has many disadvantages,such as poor selection accuracy.In order to overcome the disadvantages,this research proposed a new feature selection method based on Wiener filtering and enhanced matrix.Noise of the warning image was eliminated via the Wiener filtering.After that,gray value of pixel in the warning image was calculated via horizontal increment and vertical increment.Sum of second partial derivative of abscissa axis and ordinate axis corresponding to the gray value of each pixel was solved.According to the partial derivative,Feature matrix of the color texture was deduced.A model of feature of the color texture was obtained via the matrix.The feature selection was achieved via the model.Simulation results show that the method has excellent inhibitory effect for noise of the warning image.It can reserve information of the image preferably and select the feature of color texture under big data processing accurately.

关 键 词：大数据处理 警示性图像 颜色纹理 特征选取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]